Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "The Optical Gravitational Lensing Experiment. Red Clump Stars as a Distance Indicator"

See other formats


The Optical Gravitational Lensing Experiment. 
Red Clump Stars as a Distance Indicator^ 

A. U d a 1 s k i 

Warsaw University Observatory, Al. Ujazdowskie 4, 00-478 Warszawa, 
q ! Poland 

Cn| ' e-mail: 



We present relation of the mean /-band brightness of red clump stars on metallicity. 
Red clump stars were proposed to be a very attractive standard candle for distance de- 
termination. The calibration is based on 284 nearby red giant stars whose high quality 
spectra allowed to determine accurate individual metal abundances. High quality par- 

i- i allaxes (ov/V < 10%) and photometry of these very bright stars come from Hipparcos 

measurements. Metallicity of the sample covers a large range: —0.6 < [Fe/H] < +0.2 dex. 
We find a weak dependence of the mean /-band brightness on metallicity («0.13 mag/dex). 

What is more important, the range of metallicity of the Hipparcos sample partially 
overlaps with metallicity of field giants in the LMC, thus making it possible to determine 
the distance to the LMC by almost direct comparison of brightness of the local Hipparcos 
red clump giants with that of LMC stars. Photometry of field red clump giants in nine 
low extinction fields of the LMC halo collected during the OGLE-II microlensing survey 
compared with the Hipparcos red clump stars data yields the distance modulus to the 
I ; LMC: (m-M) LM c = 18.24±0.08 mag. 


1 Introduction 

Paczyfiski and Stanek (1998, hereafter PS) noticed that the /-band abso- 
lute brightness, Mj, of red clump (hereafter RC) giants, the intermediate 
age (2-10 Gyr) helium core burning stars, has intrinsically small dispersion 
and can be used as a "standard candle" for distance determination. The 
obvious advantage of this method is large number of these stars in stellar 
populations allowing determination of the mean brightness with statistically 
unprecedented accuracy. Moreover, RC giants are also very numerous in the 
solar neighborhood. Therefore their brightness could be precisely calibrated 
with hundreds of stars for which the Hipparcos satellite measured parallaxes 
with accuracy better than 10% (Perryman et al. 1997). It should be noted 

'Based on observations obtained with the 1.3 m Warsaw telescope at the Las Campanas 
Observatory of the Carnegie Institution of Washington. 

that RC stars are the only standard candle which can be calibrated with 
direct, trigonometric parallax measurements. Similar quality parallaxes do 
not exist for Cepheids or RR Lyr stars (cf. Fig. 2 of Horner et al. 1999). 

As in the case of any other stellar standard candle, the brightness of RC 
stars might, however, be affected by population effects: different chemical 
composition or age of the stellar system being studied, as compared to the 
local Hipparcos giants. PS argued that both dependences of (Mj) on age and 
metallicity are negligible. On the other hand, Girardi et al. (1998) claimed 
much stronger dependences based on theoretical modeling. According to 
their results (Mi) of RC stars could be different by as much as 0.5 mag in 
different environments. However, one should be aware that results of mod- 
eling are quite sensitive to the input physics and results from different types 
of evolutionary codes are not consistent each other (Dominguez et al. 1999, 
Castellani et al. 1999). While models seem to reasonably reproduce quali- 
tative properties of RC stars they cannot provide accuracy of hundredths of 
magnitude required for precise distance determination. 

The necessity of good, preferably empirical, calibration of population ef- 
fects on RC brightness became very urgent, in particular when the distance 
modulus to the LMC was determined (Udalski et al. 1998, Stanek, Zarit- 
sky and Harris 1998), supporting the "short" distance scale to that galaxy 
(~0.4 mag smaller than the "long" value of (m — M)lmc = 18.50 mag). The 
distance to the LMC is one of the most important distances of the modern as- 
trophysics because extragalactic distance scale is tied to it. Udalski (1998a) 
presented an empirical calibration of (Mi) of RC stars on metallicity sug- 
gesting only a weak dependence (0.09 ±0.03 mag/dex). The dependence of 
(Mj) of RC stars on age was studied by Udalski (1998b). Observations of RC 
stars in several star clusters of different age located in low extinction areas 
of the Magellanic Clouds showed that their (Mi) in these clusters is inde- 
pendent of age (for stars 2-10 Gyr old) within observational uncertainties of 
a few hundredths of magnitude. Recently, Sarajedini (1999) presented anal- 
ysis of a few Galactic open clusters suggesting fainter RC in older (>5 Gyr) 

In this paper we present empirical arguments which additionally support 
usefulness of the "RC stars" method - the relation of (Mi) vs. metallicity 
based on the most precise data necessary to solve the problem: high reso- 
lution and S/N spectra of nearby red giants, for which accurate parallaxes 
and photometry were measured by Hipparcos. Large range of metallicity of 
nearby RC giants partially overlapping with metallicity range of field RC 
giants in the LMC enables us to compare brightness of these stars and de- 

termine the RC distance modulus to the LMC largely free from population 

2 Observational Data 

The sample of red giant stars from the solar neighborhood comes from the 
Hipparcos catalog and consists of objects with high accuracy trigonometric 
parallaxes (cr n /ir < 10%). About 75% stars from this sample are the same 
objects which were used by PS, i.e., stars with /-band photometry. To en- 
large that data set we have also included stars which /-band magnitude was 
obtained from B — V color via very well defined correlation between B — V 
and V — I colors (c/. Fig. 8 Paczyhski et al. 1999), i.e., stars marked as type 
'H' in the Hipparcos catalog. While accuracy of the / magnitude of these 
stars is somewhat worse, it is still acceptable taking into account usually 
larger uncertainty from parallax error. 



— -1.5 

* • o __ 

• * A 

- 'J .— 

> c 

.5 — 


— .5 




Fig. 1. M/ of 284 nearby red giants with precise photometry and spectroscopy plotted as 
a function of V — I color (left panel) and metallicity [Fe/H] (right panel) . Stars of low, 
medium and high metallicity are marked by filled circles, filled triangles and open circles, 

For further analysis only the stars with [Fe/H] abundance determination 
were selected. We used results of the spectroscopic survey of McWilliam 
(1990), containing [Fe/H] determinations for 671 G and K giants based on 

high resolution (R ~ 40000) and S/N (~ 100) spectra. This is the most 
comprehensive and homogeneous data set available. Typical accuracy of 
[Fe/H] determination is about 0.1 dex. 284 objects from our photometric 
sample were cross-identified with objects in the spectroscopic survey list 
of McWilliam (1990). Left panel of Fig. 1 presents the color-magnitude 
diagram (CMD) of this sample. Comparison with Fig. 2 of PS indicates 
that full photometric sample and our 284 object sample used for further 
analysis are distributed identically in the CMD, so we did not introduce any 
significant bias when limiting Hipparcos stars to objects with spectroscopic 
data. The mean distance from the Sun of our final sample of 284 stars is 
<<i>=66 pc. It is worth noting that due to accurate Hipparcos parallaxes 
the possible Lutz-Kelker bias of the absolute magnitude of our sample is 
negligible, as shown by Girardi et al. (1998). 

The photometric data of the LMC fields were obtained during the OGLE- 
II microlensing survey (Udalski, Kubiak and Szymahski 1997). Observations 
were collected with the 1.3-m Warsaw telescope at the Las Campanas Obser- 
vatory, Chile which is operated by the Carnegie Institution of Washington 
on eight photometric nights between October 30, 1999 and November 12, 
1999. Single chip CCD camera with 2048 x 2048 pixel SITe thin chip was 
used giving a scale of 0.417 arcsec/pixel. 3 — 5 frames in the Fand /-bands 
were collected for each field with exposure time of 300 sec for both bands. 
Photometry was derived using the standard OGLE pipeline. On each night 
several standard stars from the Landolt (1992) fields were observed for trans- 
formation of the instrumental magnitudes to the standard system. The error 
of zero points of photometry should not exceed 0.02 mag. 

3 Discussion 

3.1 Hipparcos Red Clump Stars 

CMD of the analyzed sample of red giants presented in the left panel of Fig. 1 
indicates that the majority of objects are RC stars that form a very compact 
structure located in the range: 0.0 < M/ < —0.5 and 0.9 < V — / < 1.1. They 
are, however, contaminated by red giant branch stars and a blue vertical 
structure, going from Mi ~ —1.4 mag down to Mi ~0.1 mag, consisting of 
younger (more massive) red giants. 

Mi of stars is plotted against metallicity [Fe/H] in the right panel of 
Fig. 1. The distribution of stars is non- uniform with most objects within 
metallicity range of —0.3 < [Fe/H] < 0.0 dex. To investigate the relation 

of Mi of RC stars on metallicity we divided our sample into three sub- 
samples: high metallicity, [Fe/H] > —0.05 dex, medium metallicity, —0.25 < 
[Fe/H] < —0.05 dex, and low metallicity stars, [Fe/H] < —0.25 dex. Such a 
division ensures more or less uniform distribution of objects within each 
bin with the mean, median and mode values equal to (0.02,0.00,0.00), 
(-0.15,-0.14,-0.14) and (-0.39,-0.36,-0.33) dex for the high, medium 
and low metallicity bins, respectively. We determined (Mi) in all bins in 
similar manner as described in Udalski et al. (1998). Fig. 2 presents the 
histograms of distribution of Mj for each subsample with fitted Gaussian 
function representing RC stars superimposed on parabola representing back- 
ground stars. 





I I I I I I I I I 
-0.6 < [Fe/H] < -0.25 ' 

Fig. 2. Histograms of distribution of Mi of the low (top panel), medium (middle panel) 
and high metallicity sample (bottom panel) of local RC stars. Bins are 0.07 mag wide. 

It is evident from Fig. 2 that the majority of stars in the low and medium 
metallicity bins are typical intermediate age RC objects. Small dispersion 
of their Mj (<tr,c = 0.12 mag) indicates that these stars can indeed be a good 
standard candle. On the other hand the high metallicity bin has the RC 
poorly defined (ctrc = 0.26 mag). Closer examination of Fig. 1 indicates that 

most of stars from this bin (open circles) are red branch giants and objects 
located on the vertical sequence of younger giants in the RC evolution phase 
with very few stars belonging to the intermediate age RC. 

(Mj) of RC stars in the low, medium and high metallicity bins are equal 
to -0.273±0.015,-0.244±0.012 and -0.190±0.041 mag, respectively (sta- 
tistical error). The horizontal bars in the right panel of Fig. 1 show the 
range of each bin and its (Mi). The vertical bars are shown at the median 
metallicity of each bin and represent statistical uncertainty of (Mi). It is 
clear from Figs. 1 and 2 that the absolute brightness of RC stars increases 
with lower metallicity. If we consider all three bins then the slope of the 
Mi vs. [Fe/H] relation is about 0.2 mag/dex. One should, however, remem- 
ber that the high metallicity bin is poorly defined due to small number of 
metal rich RC stars of intermediate age in the solar neighborhood. More- 
over, metallicity of the most metal rich stars from McWilliam's sample is 
very likely underestimated (McWilliam 1997) making this bin additionally 
uncertain. Larger mean metallicity of this bin would lead to smaller slope 
of the Mi vs. [Fe/H] relation. Indeed, if we limit ourselves only to low and 
medium metallicity bins where the intermediate age RC stars dominate the 
linear relation becomes: 

Mi = (0.13 ±0.07) • ([Fe/H] +0.25) - (0.26 ±0.02) (1) 

Eq. (1) indicates that the dependence of Mi on metallicity is rather weak. 
The relation is in good agreement with the previous empirical determination 
by Udalski (1998a) based on comparison of RC stars with RR Lyr stars 
but this time it is based on precise measurements of numerous sample of 
individual stars, thus it is more reliable. It should be also stressed that the 
result is weakly sensitive to systematic errors, as those are very unlikely in 
the Hipparcos absolute photometry of so bright and nearby stars, and due 
to weak dependence on metallicity even large systematic metallicity error 
(which is also unlikely, Taylor 1999) would only lead to a magnitude shift 
of the order of a few hundredths of magnitude. 

3.2 LMC Red Clump Stars 

Table 1 summarizes the basic properties of field RC stars in nine fields 
around star clusters distributed in different parts of the LMC halo. The 
interstellar extinction in these directions is small, thus minimizing uncer- 
tainties of extinction-free photometry. All lines-of-sight are far enough from 
the LMC center so the reddening could be reliably determined from the 

Table 1 
Photometry of field red clump stars in the LMC 




N s , 

E(B-V) (/o> [Fe/H] 









































































COBE/DIRBE maps of Schlegel, Finkbeiner and Davis (1998). The /-band 
interstellar extinction was calculated using the standard extinction curve 
(Aj = 1.96 • E{B — V)). We assumed uncertainty of the reddening value 
equal to ±0.02 mag. 

(/) of RC stars was derived in similar manner as in Udalski et al. (1998). 
About 160-1400 field red giants were used for its determination in our fields. 
Statistical uncertainty of (/) is usually below 0.01 mag so the main contri- 
bution to the extinction free magnitude, {Io), error budget comes from the 
interstellar reddening uncertainty. Dispersion of brightness of field RC stars, 
<trc, is typically below 0.15 mag similar to the local RC stars. 

Unfortunately metallicity of the LMC field RC stars is not known so 
precisely as that of the Hipparcos local giants. Metallicity of field giants in 
our nine lines-of-sight was determined by Bica et al. (1998, herafter BGD- 
CPS) using Washington photometry. It ranges from —0.7 dex to —0.35 dex 
(Table 1) with typical error of 0.1 dex of internal determination, and the 
total error of ~ 0.2 dex. Due to large errors it is not clear if the observed 
dispersion of metallicity in different parts of the LMC is real or results from 
the uncertainty of the method. It is also not clear whether the absolute 
numbers are correct, e. g., metallicity determined using the same method for 
several LMC clusters by BGDCPS is on average by ~ 0.2 dex lower than 
spectroscopic metallicities of the LMC clusters determined by Olszewski et 
al. (1991). This point should be cleared up in the near future with direct 

spectroscopic observations of the LMC RC stars with new 8-m class tele- 

Hipparcos RC Stars 


-.5 — 


# * * 

# * 

• *a ** 



18 — 

LMC RC Stars 


Fig. 3. Mi of Hipparcos RC giants (upper panel) and (7o) of RC stars in nine fields in the 
LMC (lower panel) plotted as a function of metallicity. Asterisk in the lower panel denotes 
the mean metallicity of RC stars in the LMC and the horizontal bar spread of metallicity. 
Dotted lines indicate the range where metallicities of the local and LMC stars overlap. 
Solid thick line in the upper panel marks the derived relation of Mi on [Fe/H] (Eq. 1) 
while dashed line is its continuation resulting from Girardi's (1999a) models. Dashed line 
in the lower panel indicates possible relation for the LMC RC stars. 

Lower panel of Fig. 3 presents (Iq) of RC stars in our fields in the LMC 
as a function of BGDCPS [Fe/H]. Large asterisk indicates the mean metal- 
licity and brightness of the entire sample. Horizontal bar corresponds to 
the typical range of metallicities of field giants, ~0.5 dex, as determined by 
BGDCPS. It is worth noting that if one assumes that dispersion of metal- 
licity in the LMC fields is real and BGDCPS determinations are correct, at 
least differentially, then the observed trend of variation of (Iq) with [Fe/H] 
is similar to the local sample of RC stars. The formal fit of a straight line 
gives the slope equal to 0.21 ±0.12 mag/dex. Although one should treat this 
figure with caution it is encouraging that it is in good agreement with that 
resulting from analysis of the local RC stars. 

3.3 Distance Modulus to the LMC 

The range of metallicities among nearby RC stars is wide and covers — 0.6< 
[Fe/H] < +0.2 dex. This is a very fortunate situation, because this range 

partially overlaps at the low [Fe/H] end with metallicity of the RC stars in 
the LMC. In the upper panel of Fig. 3 Mj of Hipparcos RC stars is plotted 
as a function of [Fe/H]. Solid line marks relation of Mi vs. [Fe/H] given by 
Eq. (1). Two dotted vertical lines indicate the range where metallicities of 
the LMC and local Hipparcos RC stars overlap. We remind here that we 
have adopted BGDCPS metallicities of the LMC RC stars, and if they are 
too low, what is likely, the overlap of metallicities of both populations can 
be much larger. 

The mean metallicity of the LMC RC stars lies outside the overlap re- 
gion. Therefore, to derive (m — M)lmc we have to slightly extrapolate Mi 
vs. [Fe/H] relation given by Eq. (1). (Mi) of the Hipparcos RC giants extrap- 
olated to the metallicity of —0.55 dex is equal to (Mi) = —0.30 ±0.04 mag. 
With (Iq) = 17.94 ±0.05 mag of RC stars in our nine lines-of-sight in the 
LMC, this immediately leads to (m-M) LM c = 18.24 ±0.08 mag. 

Can this result be severely affected by extrapolation? That could only 
be possible if Mi vs. [Fe/H] relation of RC stars for metallicities in the 
range of — 0.9 < [Fe/H] < —0.5, i.e., not covered by Hipparcos stars, would 
behave extraordinarily. In particular it would have to be extremely steep 
in this range to narrow the gap between our result and the "long" distance 
modulus of (m — M)lmc = 18.50. However, this is not the case: we already 
mentioned that if dispersion of metallicity in the LMC fields as measured 
by BGDCPS is real then the slopes of Mi vs. [Fe/H] relations in the LMC 
and the local sample are similar. Also theoretical modeling of RC provides 
similar arguments. For instance, models of Girardi (1999a, Fig. 4) indicate 
that for RC stars of age 2-8 Gyr the mean slope of the Mi vs. [Fe/H] relation 
in the range of metallicities of —1.0 < [Fe/H] < —0.4 is about 0.15 mag/dex 
in excellent agreement with our empirical data (theoretical relation is plot- 
ted with dashed line in the upper panel of Fig. 3 as a continuation of our 
empirical relation). Thus large uncertainty of the distance modulus due to 
extrapolation of Eq. (1) is very unlikely. 

Possible differences of ages of both populations can be another potential 
source of uncertainty of the derived (m — M)lmc- Empirical study of this 
effect based on analysis of clusters in the Magellanic Clouds showed that 
it is practically negligible for RC stars of age within 2-10 Gyr (Udalski 
1998b). On the other hand analysis of a few Galactic open clusters by 
Sarajedini (1999) suggests that RC stars older than «5 Gyr become fainter. 
Without going into detailed discussion of these results we only note here that 
analysis of Galactic clusters requires very precise distance determination 
what is difficult and that two old clusters claimed to have fainter RC (Be39, 


NGC188) have actually very sparse population of RC stars consisting of only 
a few stars. However, despite differences which deserve further studies, both 
Udalski (1998b) and Sarajedini (1999) data show that for the age range of 
2-5 Gyr (Mj) of RC giants is constant within ±0.05 mag. The age of the 
LMC RC stars is within this range (BGDCPS) and the vast majority of 
the local RC stars are younger than 4 Gyr (Girardi 1999b). To be on the 
safe side we included to the final error budget uncertainty of ±0.05 mag for 
possible differences of age between both populations. We may conclude that 
the derived distance modulus to the LMC is largely free from population 
uncertainties. Also small interstellar extinction assures that the result is 
sound. Had there been any additional LMC extinction in the LMC fields, 
on top of that given by Schlegel et al. (1998), (m— M)lmc would be reduced 
to even smaller value. 

Acknowledgements. We would like to thank the anonymous referee 
whose critical remarks allowed us to significantly improve the manuscript. 
We are very grateful to Mr. K. Zebruh for collecting observations of the 
LMC fields. We thank Drs. B. Paczyhski, K.Z. Stanek, M. Kubiak and 
M. Szymahski for many discussions and important suggestions. The paper 
was partly supported by the grants: Polish KBN 2P03D00814 and NSF 


Bica, E., Geisler, D., Dottori, H., Claria, J.J., Piatti, A.E., and Santos Jr, J.F.C. 1998, 

Astron. J., 116, 723 (BGDCPS). 
Castellani, V., Degl'Innocenti, S., Girardi, L., Ma rconi, M., Prada Moroni, P.G., and 

Weiss, A. 1999, Astron. Astrophys., in press, astro-ph/9911432. 
Dominguez, I., Chieffi, A., Limongi, M., and Straniero, O. 1999, Astrophys. J., 524, 226. 
Girardi, L., Groenewegen, M.A.T, Weiss, A., and Salaris, M. 1998, MNRAS, 301, 149. 
Girardi, L. 1999a, MNRAS, 308, 818. 
Girardi, L. 1999b, astro-ph/9912309. 
Horner, D. et al. 1999, astro-ph/9907213. 
Landolt, A.U. 1992, Astron. J., 104, 372. 
McWilliam, A. 1990, Astrophys. J. Suppl. Ser., 74, 1075. 
McWilliam, A. 1997, AAR&A, 35, 503. 
Olszewski, E.W., Schommer, R.A., Suntzeff, N.B., and Harris, H.C. 1991, Astron. J., 101, 

Paczyriski B., and Stanek, K.Z. 1998, Astrophys. J. Letters, 494, L219 (PS). 
Paczyriski B., Udalski, A., Szymaiiski, M., Kubiak, M., Pietrzyriski, G., Soszyriski, I., 

Wozniak, P., and Zebruri, K. 1999, Acta Astron., 49, 319. 
Perryman, M.A.C. et al. 1997, Astron. Astrophys., 323, L49. 


Sarajedini, A. 1999, Astron. J., 118, 2321. 

Schlegel, D.J., Finkbeiner, D.P., and Davis, M. 1998, Astrophys. J., 500, 525. 
Stanek, K.Z, Zaritsky, D., and Harris, J. 1998, Astrophys. J. Letters, 500, L141. 
Taylor, B.J. 1999, Astron. Astrophys. Suppl. Ser., 135, 75. 
Udalski, A., Kubiak, M., and Szymariski, M. 1997, Acta Astron., 47, 319. 
Udalski, A. 1998a, Acta Astron., 48, 113. 
Udalski, A. 1998b, Acta Astron., 48, 383. 

Udalski, A., Szymariski, M., Kubiak, M., Pietrzyriski, G., Wozniak, P., and Zebruri, K. 
1998, Acta Astron., 48, 1.